Determination of Appropriate Model for Yield Gap Analysis of Rice in Guilan Province using Boundary Line Analysis Method

Introduction[1]Nowadays, identification of the yield limiting factors in the field particularly the various yield components including number of panicle per unit area, number of seeds per panicle and seed weight) is one of the most important methods to increase the production of rice. The yield gap (YG) analysis can be performed by measuring the yield related characteristics. Yield gap was estimated as the difference between actual and potential yield that has been used in various studies as an important indicator to increase the yield in crops and different areas. One of the most powerful methods to evaluate the reasons of yield potential and yield gap is boundary line analysis. The purpose of this research was to select an appropriate function for describing the relationship between yield and yield components in the Fumann plain of Guilan province. Furthermore, after selecting the superior function, the parameters of the yield and yield components were estimated  to calculate the yield gap in the region. Materials and MethodsThe present study was carried out during two cropping seasons: 2012-13 and 2013-14 in Foumanat plain (cv. ‘Tarom Hashemi’). We recorded the geographic coordinates of 53 fields. At the end of growing season (harvesting time), paddy yield and yield components (panicle number, filled grain number and 100- grain weight) were calculated in each field. The correlation coefficients between yield components and yield were studied. Segmented, quadratic and dent-like models were applied to describe the relationship between yield and yield components. Root mean square error (RMSE), determination coefficient (R2), regression simple coefficients (a & b) and coefficient of variation (CV) were used to identify the appropriate model. After selecting a superior model, the boundary line method was used to calculate yield gap and its percentage, optimum yield and optimum amount of yield components for each field. Results and DiscussionAccording to the results, a positive and significant correlation was existed between paddy yield with panicle number and filled grain number with 100- grain weight and a negative and significant correlation was existed between 100- grain weights with panicle number. Linear regression simple coefficients for all traits studied in the quadratic function and for two traits of panicles number per square meter and of filled grains number in the panicle in the segmented model were significant. Among the fitted models, segmented model has the lowest RMSE (respectively equal to 0.082 and 0.472) and coefficient of variation (equal to 1.26 and 6.39, respectively) in terms of two characteristics of panicle number and 100- grain weight and was able to describe the trend of the experimental data. In addition, dent-like model with the lowest RMSE (equal to 0.484) and coefficient of variation (equal to 6.60) used to describe the changes of filled grain number. In Foumanat plain, YG was recorded 3.63 t.ha-1with the average optimum yield and actual yield of 8.44 and 4.81 t.ha-1, respectively (40% reduction in yield). Also, the optimum amount of panicle number, filled grain number and 100- grain weight were 560, 47-83.9, and 2.18 g, respectively. ConclusionAlthough, the area of Foumanat plain in the west of Guilan province has low actual yield, there is a good potential to increase the current yield. In this study, two segmented and dent-like models were identified as superior models. The highest YG in this study was related to the number of panicles per square meter followed by the number of filled grains and the 100- grain weight. Therefore, proper crop management for improving the yield components could be an important step towards reducing the YG and increasing the yield potential in the studied area.